Economic and Employment Growth in Germany – The Sectoral Elements of Verdoorn´s Law with regional Data

CAWM-Discussion Paper No. 63

February 2013

by

Jens Oelgemöller1

Abstract

A major aspect of employment growth is discussed in relation to economic growth. This paper deals with the question as to whether the relationship between economic and employment growth, subsumed under the idiom Verdoorn´s Law, holds true at the sectoral level. For this reason, the German labor market is divided into regional functionally delineated labor markets. The employees are differentiated into sectoral affiliation, education, national status and part-time employment. The economy is split into six sectors. The labor demand function is derived from the cost-function of companies, and factor prices (interest rates and wages) are considered. It is evident that the construction sector still has intense connections to the labor market concerning output changes. This cannot be verified in the finance, insurance and service sector. Part-time work increased during the economic crisis. The elasticity to factor-prices holds true for most types of employment. It is found that, regional labor market performance is directly linked to industrial structure. The fixed an random-effects estimations used here deliver satisfying results to most investigations. However, some concerns about the results regarding characteristics of employees remain.

JEL: J21, J23, O11, R 11

Keywords: Verdoorn, sectoral growth, regional growth, employment elasticity

1 Jens Oelgemöller, Institute for Spatial and Housing Economics, University of Muenster, Am Stadtgraben 9, D- 48143 Muenster, Germany, E-Mail: [email protected]. 1. Introduction

The main targets of regional and national policies, which are omnipresent in public discussions, are sound economic growth and good labor market performance. The latter is represented either by unemployment or employment rates. A positive relationship between a prosperous economy and positive development of labour market outcome has been deemed to be undisputed. Indeed, this idea is based on the assumption of adjusting labour demand due to changes in production. The common “rule” related to this is called “Verdoorn´s law,” which states that the growth of (labour) productivity is an endogenous result of output growth – focusing particularly on the manufacturing sector. Although Verdoorn´s law refers to a study of Petrus J. Verdoorn (1949), the common interpretation considering employment growth was mainly developed by Kaldor (1966). He showed that the change in labour productivity could be measured as the difference between output and employment growth. Assuming a linear relationship of output and productivity growth with the latter being the dependent variable, this can be – considering Kaldor´s findings - transformed to a linear relationship between employment and economic growth.2

The importance of economies of scale has to be stressed. Higher production tends to result in increasing divisions of labour and productivity gains. This leads to employment growth. Traditionally, studies concerning the Verdoorn-Law are focused on the manufacturing sector. These, however, do not take into account the situation in economies with diversified economic structures. In Germany, the share of gross value added by the manufacturing sector (without construction) to the total gross value added was about 25 % in 2010. Thus, the influence of other sectors on employment development should be considered. The same holds true if talking about employment rates which can be segregated into diverse components as well. Besides sectoral affiliation, age and the human capital of employees will be of interest. The human capital requirement should vary between sectors and the production structure of sectors is far from being unique. The labor demand will differ concerning employees` characteristics and the labor intensity of the production. This aspect will be considered when controlling for factor elasticity. Assuming a production function with two input factors (labor and capital), there should be an inverse relationship to the factor prices, the so called Shephard´s Lemma. Following Flaig/Rottmann (2001), it will be shown that factor prices have significant influences. To discuss this aspect, costs for capital and labor, i.e. the interest rate and remuneration, will be included. As the economic crisis of the years 2008 and 2009 and the unexpected absence of dismissals can be interpreted as a structural break in the relationship between production and employment, the influential aspects of factor prices have to be stressed. Thus, the years of the financial crisis are represented by using a dummy for these years.3 However, this indicator will give information about the existence of a structural break, which should be important for the production sector or for atypical employment. The latter will be represented by part-time employment.

During the past decade, employment has grown more intensely than in prior booming periods in Germany (Eichhorst/Marx/Thode, 2009). Nonetheless, this development does not hold true for all

2 A formal description can be found in Kapsos (2005). 3 This approach is analogous to the dummy for the structural break in 1982/1983, used by Sögner/Stiassny (2002), who investigated the Okun-coefficient for 15 OECD-countries. 2 sectors, as can be seen in Figure 1.4 Increasing employment does occur in the sectors Financial, Insurance and services (FIS) and the Public and Private Service Sector (SERV) on the one hand, whereas employment within the sectors Production (PROD) and Construction (CONS) declined on the other.5

Figure 1: Total Employment per sector and total

The development of the total gross value added (GVA) of these sectors is shown in Figure 2. The movement of the GVA- and the employment-curve show the same tendencies. Especially the GVA in the Financial and the Service sectors increases constantly. Aside from the years 2008 and 2009, solid growth rates have taken place in the Production sector.

Figure 2: Development of Gross value Added (GVA) on sector-level and total

The growth rates for total employment and total GVA in Germany are plotted in Figure 3, showing the connection between economic and employment growth. This relationship does obviously hold true up until the financial crisis, at which point the enormous decline in economic growth is not reflected by decline in employment rates.

Figure 3: Growth rates of employment and gross value added

This paper tries to shed a diversified light on the German labor market development as the “Verdoorn Law” is taken as a basic idea and the interdependencies between economic growth and changes in employment rates are identified. To do so, the German labor market is separated into its

4 Means and Standard-Errors for the growth rates of gross value added and employment are presented in Appendix A. 5 Schanne (2006) analyzed the regional development of employment in nine industries, focusing on the reciprocal relationship between regions. 3 regional counterparts. The functionally delineated labor markets by Oberst (2012) are used as administrative borders tend not to reflect the real economic relations. While an analysis of the Verdoorn-relationship at the national level neglects the high degree of regional heterogeneity within the German labor market, the use of sole administrative regions as the German districts (NUTS 3 level, i.e. the 413 German districts “kreisfreie Städte und Landkreise”) neglects the systematic interdependency between those districts. Both approaches are inadequate areas of research for the analysis of regional labour market development as they distort the behavior of the labour force (Casado-Diaz, 2000). Using travel to work areas minimizes commuting between these regions but maximizes the commuting within and thus reduces spillover effects. Hence, the correlation between changes in a locally and economically situated labor market is concentrated at its regional level. Functionally defined regional study areas, seen as sub-national economies, are useful laboratories for examining macroeconomic theory and policy (Carlino/DeFina, 2006). They enable additional data observations within one country, which are comparable with respect to legal, political and social systems. Because working with regional data might induce special problems,6 the necessity to control for spatial components is taken into account.

The paper is structured as follows. The first section presents a brief review of the literature that addresses questions relevant to this paper; the results of basic Verdoorn-calculations are presented as well. After that, the model is expanded to control more precisely for sector specific labour demand as factor prices are included. The third part takes into account different characteristics of employment, and the fourth section provides concluding remarks.

6 The motivations for, and explanations of, spatial econometric approaches can be found in LeSage / Pace (2009); there is a special focus to this issue regarding regional employment growth in Germany in Zierahn (2012). 4 2. The Basic Model

Several studies have investigated how labour markets have reacted to employment output growth. While some studies have tried to link the reduction of unemployment rates, i.e. testing the Okun relationship, to output growth, others have focused on movements of employment rates.7 Many studies emphasize international comparisons to obtain information about the flexibility of labor markets as well as to show the influence of labor market institutions like employment protection (Flaig/Rottmann, 2009; Döpke, 2001; Kapsos, 2005). Furthermore, the employment effect of economic growth in order to reduce poverty is analyzed.8 This has been done with focus on specific countries like Botswana (Ajilore/Yinusa, 2011) Cote d`Ivoire (N`Zue, 2001), Malaysia (Jiun/Nga, 2011) or Cameroon (Besso, 2010).

National investigations have been done concerning the elasticity of employment in certain sectors, mainly in the manufacturing sector (Flaig/Rottmann, 2001). The regional perspective has become more intensified in recent years. Regional comparisons have been done for Finland (by Kangasharju/Pehkonen, 2001), the United Kingdom (Hildreth, 1988), the United States (McCombie/de Ridder, 1984) and Spain (León-Ledesma, 2000). Alexiadis/Tsadis (2006) used spatial models for Greek regions and stressed the geographical dualism concerning manufacturing agglomeration. Kangasharju/Pehkonen (2001) additionally found that the sector structure in the Finish region had a significant influence, and the private service sector a major impact. This confirms the results from León-Ledesma (2000), who analyzed the relationship between the agriculture, construction, manufacturing and services sectors, focusing on economies of scale. He found legal support for the manufacturing and construction sectors but none for agriculture. The service sector showed positive effects as well, but there were estimation problems and results should be interpreted cautiously. Kunz (2012) investigated Germany’s regional labor market, analyzing particularly regional unemployment disparities and adjustment paths in Germany. Kosfeld/Dreger (2006) solely used functionally delineated areas, focusing on spatial dependencies and sector composition.

This study deals with regional data from Germany by choosing data on NUTS III level. These regions are based on administrative borders and economical interactions between will be present. Thus, they have to be aggregated into areas where spillover effects are minimized. This can be done by developing functional areas, such as travel to work areas (TWA). By doing so, commuting within a TWA is maximized while the commuting between them is minimized. In fact, there are several methods for building these areas resulting in a different number of regions.9 In this case, the delineation by Oberst (2012) with 110 functional labour markets in Germany is chosen, as this is a good compromise between quite narrow and wide boundaries.10 The gross value, added in each sector for all administrative districts k within the TWA r, has to be aggregated, which takes the form:

∑ (1)

7 Both laws are analyzed in Herwartz/Niebuhr (2007). 8 This topic is broadly discussed by Hull (2009), Islam (2004) and Loayza/Raddatz (2010). 9 See, for example the 50 labor market regions defined by Kropp/Schwengler (2011), and the 141 German Labour Markets of Kosfeld/Werner (2012). Oberst (2012) is based on an evolutionary computational approach; Kropp/Schwengler (2011) used a graph theory approach; and, Kosfeld/Werner (2012) employed a factor analysis. For a comparison of six different labour market delineation see Kropp/Schwengler (2012). 10 The assignment of administrative districts to functional labour markets is presented in the Appendix J. 5 The same has been done for employment data. This allows for a more differentiated look at German performances and permits working with a higher amount of observations within one institutional setting in comparison to studies at the national level. Furthermore, the socio-economic factors show greater homogeneity, and production factors should be more mobile as barriers should not exist in contrast to international studies (McCombie/de Ridder, 1984). As Thirlwall (1980) states, the high level of mobility between subnational regions prevents a shortage of supply of production factors. Thus, growth is determined by demand and not by supply, which is important for model specification. Actually, the estimations conducted with 109 TWA, as the region of “Diethmarschen,” have to be removed from the sample. This became necessary as several implausible and extreme changes within regional gross value added, e.g. a 117-percentage increase within the production- sector from 2003 to 2004 (among other problems), were identified. Another data-adjustment had to take place for the region of “Altenburger Land”. Here, the amount of total gross value added was far above the sum of sectoral values in the year 1999 and was substituted by the latter.

As in common studies, the basic Verdoorn-Law is estimated by regressing output-growth on employment growth. Flaig/Rottmann (2001) added the component of dependence on factor prices and capital accumulation. The assumed substitution of labour through capital can be considered by the inclusion of the factor prices of capital (interest rate) and labour (remuneration). The rudimental expression of the Verdoorn takes the form:

(2) where E is the number of employees and GVA the gross value added in year t. By using growth rates

(g), the elasticity of the labour market on output changes is estimated. The coefficient β1 is known as the Verdoorn-coefficient and represents the marginal influence of growth on employment. The results of estimation based on equation (2) are presented in Table 1, column (A), row “total GVA”, with total employment rate as a dependent variable.

In the first model-specification, the Verdoorn-coefficient β1 for each sector s is measured:

, (3) as the regional sectoral output-growth is regressed on the regional changes of sectoral employment growth. The results can be found in column (A) in Table 1, Model (3). For all coefficients, the t-value is presented below the β-values. Additionally, the constant α is shown by the estimations R² in the cell beneath.

This simple model is expanded by adding the time-lag of economic growth to allow for adjustment processes within labor demand driven by output changes. A further extension within this model is done by employing a Fixed-Effects-model, which can be expressed by adding the Dummy ρ for each TWA r.11 This is done in accordance with the approach used by Kapsos (2005), who added Dummies for each country in his international comparison of the rate of employment growth. As annual data is used, the coefficient β2 shows the elasticity of regional and sectoral labor demand on regional and sectorial economic development, which happened a year ago; this is presented in column (B):

11 The Fixed-Effects were dominantly used, supported by the Hausman-Test, with χ2 ranges between 14.08 (FVU, model (4)) and 1437.42 (PPS, model (1)). However, the estimations have worked with random effects as well. The results and χ2are presented in the Appendix B. 6

. (4)

Finally, the differentiation between the economies of western and eastern Germany with β1 and β3 for controls in the same period and β2 and β4 for time-lagged gross value added, is done in formula (5). To separate eastern and western-data the growth-rates are interacted with the Dummy-variable Dr or (1 – Dr) respectively: 12

( ) ( )

. (5)

13 The coefficients β1-4 in formula (5) are shown in column (C).

These quite simple regressions provide interesting results. There are only little explanatory power for the sectors Agriculture (AGR), Finance, Insurance and Services (FIS) and - in model (3) - Public and Private Services (PPS). The service sectors, FIS and PPS, are the most heterogeneous and therefore difficult to classify with a single model.14 Obviously, the inclusion of time-lagged gross value added improves the model-quality as R² rises for all sectors; see column (B). This appears at a low level, especially for AGR and FIS. With respect to explanatory power, the differentiation between West- and East-Germany has a low but positive impact. It is notable that there are variations in how labor markets react to economic growth. For example, the east’s construction (CON) and PPS sectors seem to be more sensitive to economic growth than their western counterparts. Within the Trade, Transport and Hospitality (TTH) sectors it’s the other way round. The construction sector shows the highest employment elasticity to economic growth. Surprisingly, the impact of the production sector on the labor market is rather weak at first glance. Apparently, some time is required to adapt output growth when hiring employees.

Table 1: Verdoorn-Coefficients total and by sector

Effects on employment in total… (A) (B) (C) Model (2) (4) (5) West East Fixed Effects t t t-1 t t-1 t t-1

coef β1 α β1 β2 α β1 β2 β3 β4 α (t) | (R²) t-value R² t-value t-value R² t-value t-value t-value t-value R² 0.1612 0.0097 0.1662 0.1261 -0.2905 0.1575 0.1164 0.2074 0.1639 -0.2987 total GVA 14.01*** 0.167 15.04*** 9.34*** 0.235 12.95*** 7.67*** 7.90*** 5.57*** 0.239 Effects on employment per sector….. Model (3) (4) (5) 0.0568 -0.9642 0.0678 0.0436 -0.9461 0.0677 0.0524 0.0672 0.0276 -0.9465 Aggriculture (AGR) 7.74*** 0.058 9.07*** 5.69*** 0.0878 7.29*** 5.51*** 5.24*** 2.14*** 0.090 0.3714 -2.2161 0.3411 0.1776 -2.1598 0.2801 0.1273 0.4169 0.2626 -1.9587 Contruction (CONS) 23.78*** 0.366 23.34*** 12.99*** 0.4592 16.16*** 8.16*** 15.12*** 9.38*** 0.489 0.0700 -0.7152 0.0724 0.0693 -0.9083 0.0734 0.0555 0.0668 0.1093 -0.9260 Production (PROD) 9.33*** 0.082 9.94*** 7.95*** 0.1374 8.85*** 5.51*** 4.45*** 6.38*** 0.144 Trade, Hospitality, 0.2022 0.1164 0.2171 0.0651 -0.0416 0.2400 0.0766 0.1023 0.0009 -0.0322 Transportation (THT) 19.28*** 0.275 20.21*** 5.19*** 0.2945 20.61*** 5.62*** 3.94*** 0.03 0.312 Finance, Insurance, -0.0021 3.1508 0.0176 0.1397 2.6114 0.0113 0.1419 0.0344 0.1322 2.6135 Services (FIS) -0.08 0.000 0.68 5.60*** 0.031 0.37 4.88*** 0.7 2.70*** 0.031 Public and private -0.0552 1.0159 -0.0440 0.2348 0.5155 -0.0394 0.1696 -0.1085 0.5873 0.4851 services (PPS) -2.84*** 0.008 -2.43** 12.46*** 0.1438 -2.07** 8.54*** -2.40** 12.681*** 0.202 ***, **, *: Significant at 1, 5, 10%-level

12 Dr = 1 if the TWA includes at least one district belonging to former DDR, as these TWA lie in the former “Zonenrandgebiet” which were structurally weak areas in general. 13 Robust standard errors are used in all estimations (5, 6.1a, 6.1b, 6.2) here, as the Modified Wald test indicates heteroskedasticity. 14 For a similar argument, see Leon-Ledesma (2000). 7 These simple models show some relationship between employment and economic growth, but results should be interpreted cautiously. In total, the connection between output and labour does still exist, but obviously a regional labour market does not benefit from general output growth equally. The sector structure is an important factor when analyzing this economic law at the regional level as will be shown in the following. Further model modifications have to take place to get more elaborate results, which will be done in section 3.

8 3. Extended Model – Labour demand and Factor Prices

Theoretical implementation

As previously discussed, factor prices (wages and interest rates), expressed in growth rates, will be implemented. Interest rates are taken from the OECD’s Monthly Monetary and Financial Statistics (MEI), which deliver national interest rates at equal rates for every region.15 To differ between short- and long-term investments, short- (sti) and long-term (lti) interest rates can be chosen. Substituting the factors labor and capital should mainly depend on the long-term interest rate. The short-term interest rate is more applicable in cases concerning short-term adjustments in production, which would be of interest if monthly or quarterly data were available. Moreover, due to the intense drop of short-term interest rates in 200916 there is a high degree of correlation between several sectoral GVA-values and the crisis-dummy. Thus, the long-term interest rate is used next.

The implementation of wage-data at the regional and sector level is more complex. Unfortunately, sector wage developments at the regional level are not available, besides those that are within the production sector. In Table 2, model (6.1a), the development of total employment growth per region is estimated. Here, the wage rate was not separated at the sector level but at the regional level. More precisely, the variable “remuneration” (gwage_), which includes average regional income per full time employee, was chosen.17 Sectoral differentiation, as will be done in estimation (6.1b), Table 3, contains data on national average per sector (gwage_sectors). The sector ‘Production’ is an exception as information about “remuneration in production” at the regional level can be achieved and has been included into the sectorial-specific regression for production (Table 3, 6.1b). Although this variable is not directly comparable with the more general variable gwage_sector, explanatory power would be lost if this information would have been omitted.

It might be of interest to investigate the effect of sectoral growth, in addition to the effect of national growth on regional employment, as national growth would contain information about regional and sectoral interactions and reflect the overarching picture, as it is discussed most in media. But, national and sectoral economic growth is naturally highly correlated and therefore remains unconsidered in the following.18

Besides regional structural breaks (Dummy D, as presented above) effects directly connected to a certain year may occur. These time-effects could be captured by time-dummies for each year. But multicollinearity of the time-dummies is a major problem in estimations here. Therefore, the financial crisis, starting in 2008, is underlined using a Dummy C for the years 2008 and 2009. This Dummy is interacted with the regional-dummy D to show the different impacts of the crisis on western and eastern Germany whereas the time-dummies are omitted.

15 As a high mobility of capital and near perfect capital markets is assumed, interest rates without regional differentiation are quite realistic. 16 See Figure 5 in Appendix C. 17 The average income per anno (2007-2009), differentiated by sectors, are presented in Appendix D. 18 Furthermore, the correlation between changes in the long-term interest rate and national economic growth is 0.74. The high impact of national-gross value added growth on the regional unemployment rate is discussed by Oberst/Oelgemöller (2013). 9 Following Flaig/Rottmann (2001), who showed that the employment threshold, i.e. the growth rate of economic output required to hold employment constant, depends in the short run on the growth rate of relative factor prices, capital accumulation and technical change; the underlying cost-function of employers is therefore given with

( ) (a) where the variable costs Cv of production Y depend on factor prices for labour w, capital i and labor productivity P.19 Productivity can be shown as reciprocal of employment elasticity20 so that the demand of Labour L and Capital K in the short-term can be written with respect to Shepard´s Lemma (Shephard, 1954) as:

( ) < 0 (b)

( ) < 0. (c)

Cv (a) is concave in w leading to , and thus labour demand will decline if wages increase. The same holds true for the demand of capital concerning changes in interest rates. Assuming a cost function with the form

(d) the labor demand is received under consideration of (b) as:

( ) (e)

This shows labor demand´s dependency on economic output and factor prices:

and (f)

Note that demand from low-income-employees li, having an average wage level (wli) below the average wages used here, i.e. wli < w, will react positively to changes in average wages:

( ) (g)

Besides the cost-character of wages, it is often argued that higher wages lead to higher demand and therefore have positive effects on the economy and hence on employment. This dual nature of wages is debatable and is not questioned further here.21

Changes in unemployment in region r finally depend on changes of economic growth gGVA separated by sector s in region r and changes in factor prices FP, which lead to equation 6:

19 Flaig/Rottmann (2001) used a quite different cost-function, as they included prices for intermediate goods and the stock of capital. Here, however, a production function with labour and capital as input factors is chosen. The focus lies on factor prices and does not take the accumulation of capital and labor into account. Additionally, intermediate goods are omitted due to lack of data. The technical progress is also unattended as a period of ten years is investigated. The technical change would be of more interest, if longer lasting periods would be compared. 20 See Löbbe (1998) and Kapsos (2005) for an arithmetic foundation. 21 Discussion on this issue can be found in Jerger/Michaelis (2003) and Lurweg (2009). 10

, (6) where X represents a vector of control variables. The factor prices can differ between sectors and/or regions. The perfect mobility of capital leads to equal interest rates for all regions and sectors, as presumed. The time-Dummy C controlling for structural breaks during the financial crisis is one example for X.

Dealing with regional data requires controlling for spatial dependencies. It is quite possible that developments in region A have influences on developments in region B, especially if these regions are adjacent. Using functionally delineated labor markets is one way to deal with this issue. But indeed, these markets are based on administrative districts which are not functionally delimited. This might lead to imperfect areas and therefore to interdependencies between the regions. To sum it up, it might be necessary to take into account economic growth in neighboring regions (spatial-lag) and/or control for imperfect delineation (spatial error). One standard approach for checking spatial autocorrelation is to use the so-called Moran´s I, which has been done for the regional variables (employment and (sectorial) economic growth) as well as for the residuals of the estimations. The Moran’s I statistics (Cliff/Ord, 1981) as global measure of spatial autocorrelation are reported in the Appendix E and do not strongly support the necessity of using spatial-lags or spatial-error- estimations. There are few significant values, but indeed they do not occur structurally, e.g. in certain years or clearly for a certain variable. Thus, spatial-models are left out here but might be an interesting aspect in further research.22

The question as to whether using fixed or random effects is more profitable – as mentioned in section 2 – has been decided by means of the Hausman-Test. Model 6.1a23 has been done with random effects; the Hausman-Tests support random effects and fixed effects for models 6.1b and 6.2 depending on sector or employment characteristic. However, the corresponding random- (6.1a and 6.2) and fixed- (6.1b) effect-estimations are presented in the Appendix as well.

Estimation results – sectorial differentiation As the results of equations (3)-(5) confirm, the gross value added component is separated into East and West (D resp. 1-D) and time-lags are included as well. This leads to equation (6.1a), which is the sectoral expansion of equation (5). The variable remuneration is added to test for the wage- component w. The costs of capital i are represented with the long-term (lt_ir) interest rates. Wages and interest rates are included in growth rates (“g”) as changes in labor demand will be driven by changes in factor prices:

( ) ( )

( ) . (6.1a)

Table 2 presents the estimation results for sectoral undifferentiated employment growth. Besides the time-lagged coefficient of THT in eastern Germany all (significant) parameters show the expected sign. The different results for eastern and western Germany are obvious. The reagibility of employment growth to output growth is higher in the east for the sectors CONS and PPS in

22 See, for example, Oberst/Oelgemöller (2013), which uses a spatial Durban Error model with spatial lags and Kosfeld/Dreger (2006) with their spatial SUR model for spatial analysis of Okun`s and Verdoorn`s law for German regions. 23 The χ²-values are presented in the Appendix B. 11 particular. The parameters for FIS in eastern Germany are not significant in contrast to the western results. The same holds true for the crisis-dummy. This might be carefully interpreted as evidence that the economic crisis hit the western regions harder than the eastern ones.24 The positive coefficient indicates a positive employment effect during 2008/2009 without influencing economic growth. It is notable that in this model the long-term interest rate is the factor price that has an effect on employment growth. Changes in regional average wages have a negative effect, as suggested in (b).

Table 2: Sectorial Growth effect on total employment growth

Variable Coeff. z-Value AGR_west 0.01266 5.10 *** AGR_west_vp 0.00798 3.56 *** Agriculture AGR_east 0.00985 3.38 *** AGR_east_vp -0.00133 -0.49 CONST_west 0.02681 3.70 *** CONST_west_vp 0.03932 6.49 *** Construction CONST_east 0.07288 11.83 *** CONST_east_vp 0.06966 9.08 *** PROD_west 0.01721 3.70 *** PROD_west_vp 0.01279 2.20 ** Production PROD_east 0.01340 2.06 ** PROD_east_vp 0.01592 2.35 ** THT_west 0.09351 8.12 *** Trade, THE_west_vp 0.01803 1.40 Hospitality, THT_east 0.02019 1.44 Transportation THT_east_vp -0.04175 -2.30 ** FIS_west 0.01964 1.87 * Finance, FIS_west_vp 0.01611 2.14 ** Insurance, FIS_east 0.00832 0.47 Services FIS_east_vp -0.00860 -0.76 PPS_west 0.05622 2.46 ** Public and PPS_west_vp -0.00287 -0.22 private PPS_east 0.14340 6.24 *** services PPS_east_vp 0.02372 0.80 crisis west 0.48657 4.20 *** Crisis crisis east -0.07798 -0.43 gwage_ -0.08359 -2.57 ** Factor Prices interst rate 0.64172 7.92 *** _cons 0.05382 0.64 Observations = 1090 R² = 0.5779 Wald chi2(34) = 4631.21 Prob > chi2 = 0.0000 χ² =2.12 Prob > chi2 = 1.000 ***, **, *: Significant at 1, 5, 10%-level

This “atypical” labour market effect suggested by the crisis-west-dummy has been accordingly identified concerning the unemployment rate, as fewer employees had been dismissed than in previous years. The IMF (2010) took apart the cumulated changes of the unemployment rate during 2008 and 2009. For Germany, it found a predictable (using Okun´s Law) increase in the rate of unemployment. But, simultaneously, there had been an unexplained component, which reduced the unemployment rate. This component compensates for the Okun-effect, leading finally to a stable labor market during the Great recession. Both effects meanwhile are called “Germany´s Job miracle”.25

It is not likely that all sectors face the production-function mentioned above and hence won´t show additional results for the factor prices. This will be checked in the next step. Equation (6.1a) can be estimated for each sector leading to (6.1b):

24 Information about regional differentiated impact of the economic crisis can be found in Möller (2010) and Burda/Hunt (2011). 25 This idiom is referred to by Paul Krugman (2009). A detailed explanation of this “miracle” is given by Möller (2010), Schneider (2012) and Boysen-Hogref/Groll (2010). 12

( ) ( )

( ) . (6.1b)

Employment and economic growth is included at the sectoral level analogous to (5), Table 1. Results are presented in Table 3. The model is estimated with fixed effects, excluding the construction sector. Here, random effect estimation is evident:26

Table 3: Sectorial GVA growth rate on sectorial employment growth rates

Trade, Hospitality, Finance, Insurance, Public and private Agriculture Construction Production Transportation Services services

(Fixed Effects) (Random effects) (Fixed Effects) (Fixed Effects) (Fixed Effects) (Fixed Effects) Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value AGR_west 0.07287 7.3 *** AGR_west_vp 0.04466 5.09 *** Agriculture AGR_east 0.07036 6.33 *** AGR_east_vp 0.00217 0.18 CONST_west 0.22647 9.01 *** CONST_west_vp 0.09101 6.76 *** Construction CONST_east 0.42631 13.05 *** CONST_east_vp 0.33098 9.47 *** PROD_west 0.07548 4.42 *** PROD_west_vp 0.05799 4.14 *** Production PROD_east 0.07876 3.92 *** PROD_east_vp 0.10352 5.3 *** THT_west 0.25546 12.51 *** Trade, THT_west_vp 0.09466 5.78 *** Hospitality, THT_east 0.13182 5.24 *** Transportation THT_east_vp 0.03160 0.97 FIS_west -0.00599 -0.2 Finance, FIS_west_vp 0.12657 3.51 *** Insurance, FIS_east 0.02946 0.94 Services FIS_east_vp 0.11256 3.44 *** PPS_west -0.05908 -1.06 Public and PPS_west_vp 0.16611 2.24 ** private services PPS_east -0.13785 -3.8 *** PPS_east_vp 0.57379 12.58 *** crisiswest 2.07137 5.27 *** 1.13623 3.91 *** 1.75046 8.77 *** 0.99806 12.03 *** 1.18339 4.56 *** 0.18606 0.99 Crisis crisisost 3.71892 6.17 *** -1.13280 -2.54 ** 2.11360 4.53 *** 0.98657 5.87 *** 2.26451 5.76 *** 0.16464 0.64 gwage_ -0.06633 -0.77 ** -0.03579 -0.45 -0.02458 -0.29 -0.40305 -8.97 *** 1.35505 14.20 *** 0.11298 1.26 Factor Prices interest rate 0.43139 1.02 1.34794 7.28 *** 1.75564 14.45 *** 0.90143 8.56 *** -0.19512 -1.10 0.17607 2.74 *** _cons -1.37387 -29.37 *** -1.73203 -13.91 *** -1.03688 -6.65 *** 0.19388 6.00 *** 0.16618 0.78 0.41119 2.91 *** Observations 1090 1090 1090 1090 1090 1090 R² 0.1510 0.5148 0.3056 0.4473 0.1409 0.2172 χ² 39.7300 12.5100 41.2100 31.8600 46.7300 119.7200 Prob > chi2 0.0000 0.1298 0.0000 0.0000 0.0000 0.0000 F (8,108)|Wald chi2 (8) 34.91 1906.6 79.02 74.97 36.51 33.64

The separation of employees into their sectoral affiliation results in positive relations of sectoral economic growth to sectoral employment growth in almost all cases. It is only the coefficient of growth in the public service sector in eastern Germany that has obviously negative effects on employment in that sector; this would require further investigations. As the connection between sectoral output and sectoral employment is greater than overall employment, the coefficients show higher values than in 6.1a. Especially the construction sector has notable values – again with higher elasticity in eastern Germany. In the sector FIS a time-lag is required to identify the positive growth effect. Note, the explanatory power of this model is rather weak for the FIS sector, as the R² = 14,09 is comparatively low. The same holds true for the AGR. Besides the sector PPS the crisis-dummies are significant, with a surprising negative sign for eastern Germany in the construction sector. This indicator has by far the highest amount of the growth-coefficients, again indicating the extraordinary behavior of the labour market in 2008/2009. Regarding the factor prices, signs are expected to be, with FIS as exception. This positive effect of growing average wages here stresses the necessity of

26 Corresponding random and fixed effects estimations are presented in Appendix G. 13 model specification in further research.27 It might be argued, that growing wages induces migration. As high-skilled workers are more mobile in general and the skill level in the FIS-sector is comparatively high, employees might be attracted by rising wages.28

Estimation results here are quite satisfying, as coefficients are as assumed. Nonetheless, there is space for further model improvements. Besides quarterly data more specific information about cost- and production function at the sectoral level should be implemented. Regional characteristics could be more precise, too. The differentiation between Eastern and Western Germany is one possibility, but additional aspects, like infrastructure or degree of agglomeration29, could be of interest.30 An important element is the lag regarding regional sectoral wage-data. As Blien/Suedekum (2007) found, differences in regional wage levels in neighboring regions have an influence on a region´s level of employment.

Estimation results – Employees differentiation Besides the necessity to differentiate between sectors concerning economic growth, it is also useful to shed light on different forms of employees and employment (e). Using equation (6), the normal employment growth rate is substituted with the following variables Ee, each in growth rates and each at the regional level for district k31:

- foreign workers, measured as employed foreign worker per 100 foreigners able to work, - employees with low education per 100 inhabitants in working age, - employees with high education per 100 inhabitants in working age, - employees in part-time per 100 inhabitants in working age.

This is expressed in estimation (6.2)

( ) ( )

( ) . (6.2)

Unfortunately, information about mixtures of these types of employment, e.g. what kind of education foreign workers have and if they are part-time workers, is not available for the regional and sectoral levels. Foreign workers are assumed to have a relatively low or inadequate level of education. Thus, foreign and less educated workers face relatively high responses to growth within the construction and agricultural sectors as the required skill level tends to be rather low here. Low- income groups thus benefit from a rising average wage level (see (g)), as they have relative cost advantages in this case, because their wage demand would be below the average wage rate illustrated by the variable (gwage_). Part-time workers should be found in most sectors. This form of atypical employment increased significantly during the crisis, as shown in Figure 4, which presents the share of part-time workers as an average for all districts, and as an average for those 10 districts with the highest and lowest rates. The extreme increase in 2008 is obvious especially in those districts with a “natural” high rate of part-time workers.

27 The results are in line with Henderson (1997), who argues that this result might reflect “the absence of a specific wage variable for highly skilled workers” (Henderson (1998), p. 468). An alternative approach to estimating regional wage levels is presented in Blien et al. (2006). 28 The relation between wage flexibility and mobility of workers is discussed in Topel (1986). 29 Agglomerations seem to have a strong and dynamic growth, as found in Dauth (2010). 30 Further aspects are discussed in Zierahn (2011). 31 These variables have been aggregated analogously to (1). 14 Figure 4: Part-time workers (% of inhabitants in working age)

The regression-results are presented in Table 4.32 The estimation for highly-educated workers has been done with random effects, the other ran with fixed effects. The model fitting with regard to the R² is fine and most signs are as expected.

Table 4: Estimation results: Differentiation of employment

foreign workers low educated workers high educated worker part-time workes Variable (fixed effects) (fixed effects) (random effects) (fixed effects) Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value AGR_west 0.02110 1.78 * 0.00292 2.30 ** -0.01345 -1.83 * 0.00666 0.51 AGR_west_vp 0.00991 0.93 0.00155 1.38 0.01201 1.37 0.13809 8.93 *** Agriculture AGR_east 0.02304 0.76 0.00769 7.16 *** -0.01146 -1.90 * 0.03207 1.85 * AGR_east_vp 0.06429 1.20 0.00177 1.47 0.00399 0.58 0.02542 1.72 * CONST_west 0.15263 4.68 *** 0.00070 0.24 0.07308 4.67 *** -0.11249 -2.70 *** CONST_west_vp 0.07438 2.92 *** 0.00491 1.94 * 0.02935 1.79 * -0.11893 -3.12 *** Construction CONST_east 0.60729 5.52 *** 0.02690 6.52 *** 0.11719 5.87 *** 0.15890 3.90 *** CONST_east_vp -0.18148 -0.93 0.00909 2.59 ** 0.07916 2.86 *** 0.19750 3.91 *** PROD_west 0.03805 1.64 -0.00324 -1.89 * -0.00186 -0.16 0.18979 4.87 *** PROD_west_vp 0.04807 2.17 ** 0.00372 2.31 ** 0.02733 2.56 ** 0.09590 3.55 *** Production PROD_east -0.10754 -1.37 -0.00255 -0.99 -0.01323 -0.88 0.05157 1.32 PROD_east_vp 0.08262 0.91 -0.00032 -0.14 -0.00801 -0.57 0.07589 2.61 ** THT_west 0.13874 3.01 *** -0.00043 -0.09 0.03547 1.62 0.60908 7.81 *** Trade, THT_west_vp 0.09893 3.08 *** 0.03009 5.98 *** 0.04886 1.69 * -0.14650 -2.04 ** Hospitality, THT_east -0.17826 -0.65 -0.03167 -3.69 *** -0.03528 -1.31 0.20968 3.04 *** Transportation THT_east_vp -0.01384 -0.07 -0.00134 -0.10 -0.05237 -0.93 -0.14079 -1.19 FIS_west 0.10009 2.16 ** 0.00730 1.49 0.02365 1.05 0.02264 0.46 Finance, FIS_west_vp -0.03488 -1.27 -0.00063 -0.13 0.00093 0.04 0.06356 1.42 Insurance, FIS_east -0.01982 -0.07 0.00059 0.09 -0.01318 -0.45 0.04556 0.67 Services FIS_east_vp 0.13981 0.86 0.00491 0.95 -0.06342 -2.12 ** -0.09835 -1.67 * PPS_west 0.04479 0.78 0.00939 1.34 0.04107 1.07 -0.31156 -2.70 *** Public and PPS_west_vp -0.03338 -0.72 0.01191 1.95 * 0.07855 2.14 ** -0.03285 -0.53 private PPS_east -0.01641 -0.03 0.00873 0.70 -0.20271 -1.58 -0.11234 -0.52 services PPS_east_vp -0.55329 -2.03 ** 0.01784 1.59 -0.02581 -0.41 -0.47579 -3.60 *** crisiswest 4.03664 8.24 *** 0.92097 14.48 *** 1.54917 5.54 *** 21.54280 21.40 *** Crisis crisisost 2.19934 1.25 1.32754 9.75 *** 1.41330 2.59 ** 7.33504 5.99 *** gwage_ 0.13592 0.70 0.10344 6.72 *** -0.03082 -0.43 0.59774 3.88 *** Factor Prices interest rate 2.69041 4.78 *** 0.13353 3.89 *** 1.10508 5.77 *** 2.98264 5.89 *** _cons -0.48650 -0.80 13.22349 373.90 *** 2.47672 15.80 *** 2.46334 6.62 *** Observation 1090 1090 1090 1090 R² 0.2849 0.6585 0.5021 0.7308 χ² 43.9400 52.1900 39.8100 45.9500 Prob > chi2 0.0294 0.0031 0.0688 0.0176 F(28.108) 28.07 45.34 21.51 157.28 ***, **, *: Significant at 1, 5, 10%-level

Foreign workers benefit most from growth within the construction sector with higher elasticity evident in eastern Germany. This observation holds true for all estimations in this study. Positive effects can also be identified for the sectors production (time-lagged in western Germany), THT in Western Germany and FIS. The negative coefficient for the time-lagged coefficient in PPS in Eastern Germany is however unexpected. The high value of the Dummy variable “crisiswest” (4.03) indicates a labor market situation for foreign-workers which has been far more robust than it could have been assumed. The reagibility to factor prices shows a significant positive and high (2.69) amount. While

32 Again, fixed- and random effects are used for these estimations. The analogue results using fixed- and random-effects are presented in the Appendix H. 15 the sign is expected, the level is surprising and should be part of further investigations. Low-educated workers benefit from growth in the agriculture- and construction-sector with Eastern German regions showing higher elasticity. Negative indicators are seen in the production sector (west) and THT (east). The FIS-sector does not react to the hiring or firing of low educated workers as a response to output changes. The PPS-sector has a positive but low significant coefficient for the time-lagged parameter in Western Germany. This group of employees was hit by the economic crisis to a lesser extent than expected. As assumed in (g), rising average wages has positive effects on the labor demand for less-educated and, therefore, comparatively cheap workers, as their wages lie below average. The extreme value of the constant (13.22), however, raises doubts about the suitability of this calculation approach. The same holds true for highly educated workers, as negative signs appear for the sectors agriculture (low significance) and FIS. Especially the latter is unexpected; the finance sector generally requires workers with comparably high levels of education. Here, there is either no reaction to output growth or a negative one. Positive effects for the sectors Construction, THT, PPS and the crisis-dummies suit well on the other hand. The labor market situation for highly educated workers obviously does not depend on the average wage level, but is instead positively correlated to the interest rate.

The results for part-time worker are ambiguous. The agricultural sector shows positive results, as well as the construction sector in Eastern Germany and the production sector in Western Germany. In sector THT, the impact of output growth on part-time employment without time-lag is relatively high in Western (0.61) and Eastern (0.21) Germany, which is explainable by the structure of the hospitality sector. The negative sign of the parameter for the time-lagged Western Germany coefficient, however, is rather atypical. The same holds true for results in FIS (east, time-lag) and PPS (west and time-lagged east). The construction sector reveals negative signs in western Germany as well. It might be argued that part-time workers turned into full time employees when the sectors faced economic growth. This, however, cannot be proved with the data set used here. The effect of factor prices is similar to low educated employees, as the labour demand for part-time-workers increases, if the factor labor and capital become more expensive on average. The extreme value for the Dummy “crisiswest” indicates the enormous importance of atypical employment as answer to the recession in 2008/2009. But indeed, the coefficient of 21.54 here might indicate inappropriate model-fitting. While multicollinearity cannot be identified,33 this special group of employment should be part of more intense investigations, as sector affiliation, skill level and further aspects are highly relevant for this group. This requires a more complex data-set, which is not available for this study.

The estimations concerning different characteristics of employees do not show satisfying results in general. Most coefficients are assumed in the theoretical foundation, but obviously the chosen data set and methodology do not provide appropriate estimations. One important improvement would be to take into account intermixtures of employment information. Furthermore, processes for periods of less than a year, i.e. quarterly data, would deliver more detailed insights. Specifically, the seasonable impacts in the sectors of agriculture, construction and hospitality might become clear. Thus, time-lagged data would show adjustment processes more precisely. Note that structural variables might influence labor market effects of different employment groups as well. Increasing the share of high educated workers in the labor market positively affects the less educated in particular (Blien et al., 2006; Bauer, 1998).

33 The variance inflation factor (VIF) is presented in Appendix I. 16 4. Conclusion

The rate of employment growth is a major factor in labor market policy and discussions. To understand the relationship between economic output and how the regional labor markets react, several aspects have to be taken into account. First, the suspected correlation at the national level, which is more general, does not necessarily hold true on the regional scale. Differences can be identified by separating the common value of economic growth into its sectoral components. Some sectors, like construction and production, do react more strongly to labor markets and output growth than e.g. the financial and service sectors. The impact of the labor market on the latter does not seem to depend on its regional economic growth directly. Thus, regional economic structure will have an influence on employment elasticity. Labour markets in regions with a dominating financial sector might not have a high business cycle dependency, whereas the labor market in regions with a dominating second sector will be more sensitive. On the other hand, stimulating the regional labour market in the short term will be most successful in the construction sector, as it is very sensitive to changes in output growth. But the direct effect is not prolonged (see also Blien et al., 2006). This is evident in Spain, where a breakdown in the construction sector has had directly negative impacts on the labour market.

When investigating the sectorial impact on the labor market, the inhomogeneity of employees has to be considered as well. Skill requirement differs between sectors. The cost function used here regards the average demand for labor and thus a mean employee. It is not tailored to different skill levels and further characteristics. Thus, the results presented here confirm the structural break due to the financial crisis and the different reagibility to factor prices. As several parameters show unexpected results, the suitability of the model appears to be unsatisfactory. Here, some model and data specification are required to deliver more substantial results. Regarding sectorial differentiation, improvements will be generated by using quarterly data. The necessity to choose spatial econometric approaches has been discussed and left out here as the global spatial indicator “Moran´s I” does not confirm the use of spatial lag or spatial error processes. This, however, might be another aspect for further research.

17 Refrences

Ajilore, Taiwo and Olalekan Yinusa (2011): An analysis of employment intensity of sectorial output growth in Botswana. Southern African Business Review, Volume 15. No. 2, pp. 26-42.

Alexiadis, Stilianos and Dimitrios Tsagis (2006): Reassessing the validity of Verdoorn´s law under conditions of spatial dependence: a case study of the Greek regions. In: Journal of Post Keynesian Economics, Vol. 29. No. 1, pp.: 149-175.

Besso, Chrisophe Raoul (2010): Employment intensity of growth and its macroeconomics determinants. MPRA Paper No. 35615. http://mpra.ub.uni-muenchen.de/35615/.

Bauer, Thomas (1998): Arbeitsmarkteffekte der Migration und der Einwanderungspolitik. Physica, Heidelberg.

Blien, Uwe, Jens Suedekum and Katja Wolf (2006): Local employment growth in West Germany: A dynamic panel approach. Labour Economics, Vol 13, pp. 445-458.

Blien, Uwe and Jens Suedekum (2007): Stimulating employment growth with higher wages? A new approach to addressing and old controvery. In Kyklos, Vol. 60 (3), pp. 441-464.

Boeri, Tito and Herbert Bruecker (2011): Short-time work benefits revisited: some lessons from the Great Recession. Economic Policy, Volume 26, Issue 68, pages 697–765, October 2011.

Boysen-Hogrefe, Jens and Dominik Groll (2010): The German Labour market miracle. National Institute Economic Review, October 2010, vol. 214, no. 1, pp. R38-R50.

Burda, Michael C., Jennifer Hunt, Michael W. L. Elsby and John Haltiwanger (2011): What Explains the German Labor Market Miracle in the Great Recession? Brookings Papers on Economic Activity, Spring 2011, pp. 273-335.

Carlino, G., Defina, R., 2006. Macroeconomic Analysis using regional data: an application to monetary policy. In: R. Arnott and D. McMillen (Eds.), A Companion to Urban Economics, Oxford: Blackwell.

Casado-Díaz, Jose M. (2000): Local Labour Market Areas in Spain: A Case Study, Regional Studies, 34:9, 843-856.

Cliff, Andrew D. and Keith Ord (1981): Spatial Processes: Models & Applications. Pion, London.

Dauth, Wolfgang (2010): Agglomeration and regional employment growth. IAB-Discussion Paper, 7/2010.

Döpke, Jörg (2001): The "Employment Intensity" of Growth in Europe. Kieler Arbeitspapiere 1021. https://www.econstor.eu/dspace/bitstream/10419/17746/1/kap1021.pdf.

Eichhorst, Werner, Paul Marx, Eric Thode (2009): Arbeitsmarkt und Beschäftigung in Deutschland 2000-2009 - Beschäftigungserfolge bei zunehmender Differenzierung. IZA Research Report No. 22

18 Flaig, Gebhard and Horst Rottmann (2001): Input Demand and the Short- and Long-Run Employment Thresholds: An Empirical Analysis for the German Manufactoring Sector. In: German Economic Review 2 (4), pp. 367-384.

Flaig, Gebhard and Horst Rottmann (2009): Labour Market Institutions and the Employment Intensity of Output Growth. Jahrbücher für Nationalökonomie und Statistik, 2009, pp. 22-35.

Herwartz, Helmut and Annektrin Niebuhr (2007): Determinants of the (un)employment intensity of growth: Evidence from a cross section of EU regions. ttp://www.uni- graz.at/socialpolitik/papers/Niebuhr.pdf.

Hildreth, Andrew (1988/89): The ambiguity of Verdoorn´s Law: a case study of the British regions. In: Journal of Post Keynesian Economics, Vol. 11, No. 2, Winter 1988-1989, pp. 279-294.

Hull, Katy (2009): Understanding the Relationship between Economic Growth, Employment and Poverty Reduction. In: Promoting Pro-Poor Growth: Employment. Paris: OECD, 2009, 99. 69-94.

IMF (2010): World Economic Outlook.

Islam, Rizwanul (2004): The Nexus of Economic Growth, Employment and Poverty Reduction: An Empirical Analysis. ILO, Issues in Employment and Poverty, Discussion Paper No.14.

Jerger, Jürgen and Jochen Michaelis (2003): Wage hikes as supply and demand shock. Metroeconomica Vol. 54 (4), pp. 434-457.

Jiun, Chan Yu and Janice L. H. Nga (2011): Economic Growth, Employment Elasticity and Poverty: A Case of Malaysia. Prosiding Perkem VI, JILID 1 (2011), pp. 468 – 481.

Kaldor, Nicholas (1966). Causes of the Slow Growth in the United Kingdom. Cambridge: Cambridge University Press.

Kangasharju, Aki and Jaakko Pehkonen (2001): Employment-Output link in Finland: Evidence from regional data. Finish Economic Papers, Vol. 14, No. 1, pp. 41-50.

Kapsos, Steven (2005): The employment intensity of growth: Trends in macroeconomic determinants. Employment Strategy Papers, 2005/12, International Labour Office, Geneva.

Kosfeld, Reinhold, Christian Dreger (2006): Thresholds for employment and unemployment: A spatial analysis of German regional labourmarkets, 1992,-2000. Papers in Regional Science, Volume 85, No. 4.

Kosfeld, Reinhold, Alexander Werner (2012): Deutsche Arbeitsmarktregionen – Neuabgrenzung nach den Kreisgebietsreformen 2007–2011. In: Raumforschung und Raumordnung, February 2012, Volume 70, Issue 1, pp 49-64.

Kropp, Per, Barbara Schwengler (2011): Abgrenzung von Arbeitsmarktregionen * ein Methodenvorschlag. In: Raumforschung und Raumordnung, Jg. 69, H. 1, S. 45-62.

Kropp, Per, Barbara Schwengler (2012): Vergleich von funktionalen Arbeitsmarktabgrenzungen in Deutschland. In: Sozialer Fortschritt, Bd. 61, H. 1, S. 11-20.

19 Krugman, Paul (2009): Free to Lose. Published: November 12, 2009 http://www.nytimes.com/2009/11/13/opinion/13krugman.html

Kunz, Marcus (2012): Regional unemployment disparities in Germany. An empirical analysis of the determinants and adjustment paths on a small regional level. Diss. IAB-Bibliothek 331, Bertelsmann, Bielefeld.

León-Ledesma, Miguel A. (2000): Economic Growth and Verdoorn´s Law in the Spanish Regions, 1962-91. In: International Review of Applied Economics, Vol. 14, No. 1, pp. 55-69.

LeSage, James and R. Kelley Pace (2009): Introduction to Spatial Econometrics. Chapman & Hall/CRC, Boca Raton, Florida.

Loayza, Norman V. and Claudio Raddatz (2010): The composition of growth matters for poverty alleviation. In: Journal of Development Economics, Volume 93, Issue 1, pp. 137–151.

Löbbe, Klaus (1998): Sectoral Employment Elasticities in Germany. In: Addison, John T., Paul J.J. Welfens (eds.): Labor markets and Social Security. Springer, Heidelberg, pp. 119-148.

Lurweg, Maren (2009): Dogmengeschichtliche Entwicklung der Kaufkrafttheorie der Löhne, in: Vierteljahrschrift für Sozial- und Wirtschaftsgeschichte, Band 96, Heft 4, S. 511-547.

McCombie, J.S.L. and J.R. de Ridder (1984): “The Verdoorn Law controversy”: some new empirical evidence using U.S. State data. In: Oxford Economic Papers 36, 268-284.

Möller, Joachim (2010): The German labor market response in the world recession – de-mystifying a miracle. Journal for Labour Market Research, Volume 42, Number 4.

N’Zué Félix Fofana (2001): Employment and Economic Growth in Côte d’Ivoire: Analysis of Structural Determinants. African Development Review, Vol. 13 No. 1.

Oberst, Christian A. (2012) A Methodology Approach to Delineate Functional Economic Market Areas with an Iterative Three-Step Spatial Clustering Procedure. CAWM Discussion Paper No. 55.

Oberst, Christian A., Jens Oelgemöller (2013): Economic growth and the regional labor market development in German Regions: Okun's law in a spatial context. Forthcoming.

Schanne, Norbert (2006): Employment Growth in German Regions-a Spatial Econometric approach. spatial Econometrics Workshop, Rome. http://doku.iab.de/externe/2006/k060616f02.pdf.

Schneider, Hilmar (2012). Wie nachhaltig ist das deutsche Jobwunder? Eine Reformbilanz. IZA Standpunkt Nr. 51.

Shephard, Ronald W. (1954): Cost and production functions. Princeton University Press, 1953.

Sögner, L.; Stiassny, A. (2002): An analysis on the structural stability of Okun´s Law – a cross-country study. Applied Economics, 2002 (14), pp. 1775-1787.

Thirlwall, A. P. (1980): Regional Problems are ´balance payments` problems. In: Regional Studies, Vol. 14, pp. 419-425.

20 Topel, Robert H. (1986): Local Labor Markets. In: Journal of Political Economy, Vol. 94, No. 3, Part 2: Hoover Institution LaborConference (Jun., 1986), pp. S111-S143.

Verdoorn, Petrus J. (1949): On the Factors Determining the Growth of Labor Productivity. In: L. Pasinetti (ed.), Italian Economic Papers, Vol. II, Oxford: Oxford University Press, 1993.

Zierahn, Ulrich (2012): The importance of spatial autocorrelation for regional employment growth in Germany. Jahrbuch der Regionalwissenschaften, Vol. 32: pp 19-43.

21 Appendix A

The average development of total and sectorial growth of output and employment can be seen in Table 5.

Table 5: Descriptive statistics

g.GVA g.EMPL Mean Std. Err. Mean Std. Err. total 1.743 0.093 0.291 0.041 Agriculture -1.440 0.465 -1.046 0.117 Construction -0.390 0.187 -2.361 0.123 Production 1.123 0.292 -0.637 0.078 Trade, Hospitality, Transportation 1.557 0.123 0.431 0.052 Finance, Insurance, Services 3.059 0.114 3.144 0.095 Public and private services 2.205 0.076 0.894 0.049

22 Appendix B Estimation of model 2-5 with random effects and χ² of Hausman-tests.

Effects on employment in total ot sector… (A) (B) (C) Model (2) | (3) (4) (5) Random Effects West East t t t-1 t t-1 t t-1

coef β1 α β1 β2 α β1 β2 β3 β4 α (t) | (R²) z-value R² z-value z-value R² z-value z-value z-value z-value R² 0.1655 0.0023 0.1683 0.1304 -0.3041 0.1789 0.1548 0.1283 0.0488 -0.3021 total GVA 14.460 0.167 15.37 9.79 0.235 14.7 10.53 5.19 1.89 0.216 0.0586 -0.9615 0.0697 0.4534 -0.9433 0.0673 0.0543 0.0745 0.0303 -0.9440 Aggriculture (AGR) 7.990 0.058 9.36 5.96 0.0878 7.22 5.69 5.87 2.34 0.090 0.4141 -2.1990 0.3590 0.2054 -2.1420 0.2803 0.1299 0.4322 0.3127 -1.8972 Contruction (CONS) 24.210 0.366 25.14 15.73 0.4586 16.69 8.58 15.91 12.05 0.487 0.0753 -0.7212 0.0761 0.7460 -0.9270 0.0779 0.0619 0.6667 0.1094 0.9385 Production (PROD) 10.080 0.082 10.54 8.66 0.1373 9.42 6.2 4.53 6.83 0.144 Trade, Hospitality, 0.2241 0.0820 0.2387 0.0928 0.1326 0.2626 0.1060 0.1180 0.0215 -0.1229

Transportation (THT) 21.130 0.275 22.73 7.70 0.2927 23.52 8.35 4.98 0.85 0.310 Economicgrowth in…. Finance, Insurance, 0.0099 3.1140 0.0186 0.1437 2.5947 0.0235 0.1597 0.0127 0.1050 2.5824 Services (FIS) 0.390 0.002 0.75 6.02 0.031 0.83 5.98 0.3 2.61 0.030 Public and private -0.0159 0.9294 -0.0204 0.2532 0.4262 0.0208 0.2273 -0.2647 0.4258 0.4283 services (PPS) -0.820 0.008 -1.14 13.62 0.1422 1.12 11.73 -6.48 10.25 0.169

(A) (B) (C) Hausman Test χ² χ² χ² Prob>chi2 Prob>chi2 Prob>chi2 11.46 11.95 86.47 total GVA 0.001 0.003 0.000 72.06 3.18 -33.18 Aggriculture (AGR) 0.000 0.204 Contruction 150.16 31.91 25.10 0.000 0.000 0.000 Production 67.25 14.18 29.87 0.000 0.001 0.000 Trade, -192.32 57.98 54.21 0.000 0.000 Finance, 2.96 1.92 14.08 0.085 0.384 0.007 Public and 1437.42 28.24 109.6 0.000 0.000 0.000

23 Appendix C

Figure 5: Long-term and short-term interest rate

Source: OECD, Monthly Monetary and Financial Statistics (MEI)

24 Appendix D

150.00

125.00

100.00

75.00

50.00

25.00

0.00 Agriculture Construction Production Trade, Finance, Public and Grossincome peranno (Average = 100) Hospitality, Insurance, private Transportation Services services

2007 2008 2009

25 Appendix E

Moran´s I of the variables

year I p-value* year I p-value* 2000 -0.101 0.071 2000 -0.010 0.496 2001 -0.043 0.293 2001 0.046 0.182 2002 0.014 0.355 2002 -0.077 0.135 2003 0.014 0.354 2003 0.053 0.156 2004 -0.067 0.176 2004 0.006 0.401 2005 -0.022 0.421 2005 0.033 0.243

Employment 2006 0.058 0.134 2006 0.038 0.223

2007 -0.049 0.262 value Gross added 2007 -0.019 0.439 2008 0.07 0.102 2008 0.054 0.157 2009 0.051 0.169 2009 -0.037 0.325 2000 -0.044 0.280 2000 0.004 0.414 2001 -0.007 0.485 2001 -0.026 0.393 2002 0.062 0.111 2002 0.141 0.008 2003 -0.152 0.010 2003 -0.131 0.025 2004 0.125 0.015 2004 -0.072 0.158 2005 -0.051 0.248 2005 -0.029 0.373 2006 -0.01 0.495 2006 0.023 0.301

2007 0.017 0.330 2007 0.03 0.263 EmploymentTrade,

Employmentagriculture 2008 -0.011 0.486 2008 0.069 0.101 2009 -0.028 0.381 Hospitality,Transportation 2009 0.01 0.377 2000 -0.063 0.194 2000 -0.145 0.015 2001 -0.023 0.414 2001 -0.095 0.084 2002 -0.066 0.184 2002 -0.04 0.313 2003 -0.065 0.185 2003 -0.026 0.391 2004 -0.062 0.200 2004 -0.139 0.019

2005 -0.02 0.428 2005 -0.043 0.293 Services 2006 -0.101 0.071 2006 -0.093 0.090 2007 0.091 0.054 2007 0.082 0.061

Employmentconstruction 2008 -0.002 0.456 2008 0.017 0.335 EmploymentFinance, Insurance, 2009 -0.042 0.299 2009 0.053 0.156 2000 -0.045 0.284 2000 0.012 0.357 2001 -0.055 0.231 2001 0.014 0.355 2002 -0.023 0.415 2002 -0.095 0.079 2003 0.072 0.095 2003 0.001 0.436 2004 -0.034 0.345 2004 -0.058 0.217

2005 -0.077 0.138 2005 0.024 0.296 services 2006 0.01 0.379 2006 -0.093 0.086

2007 0.03 0.264 2007 0.042 0.199 Employmentproduction

2008 -0.04 0.311 2008 0.06 0.127 EmploymentPublic and private 2009 0.114 0.025 2009 -0.04 0.308 * p-values below 0.05 or below 0.1 are highlighted

26

Moran´s I of the Estimation´s residuals 6.1a, 6.1b and 6.2

Model 6.1a Model 6.2 year I p-value* year I p-value* 2000 -0.045 0.276 2000 -0.010 0.496 2005 -0.017 0.453 2001 0.032 0.236 2001 0.015 0.347 2006 0.068 0.109 2002 0.072 0.083 2002 -0.046 0.276 2007 -0.046 0.278 2003 -0.121 0.034 2003 0.027 0.282 2008 0.056 0.147 2004 0.137 0.009 2004 -0.026 0.396 2009 0.027 0.281 2005 -0.058 0.214

2006 -0.020 0.429 foreignworkers 2007 0.060 0.122 Model 6.1.b 2008 -0.012 0.481 year I p-value* year I p-value* 2009 -0.036 0.334 2000 -0.045 0.276 2000 -0.020 0.434 2000 -0.013 0.473 2001 0.032 0.236 2001 0.050 0.168 2001 0.024 0.296 2002 0.072 0.083 2002 0.085 0.065 2002 -0.068 0.173 2003 -0.121 0.034 2003 -0.116 0.043 2003 -0.078 0.134 2004 0.137 0.009 2004 -0.050 0.256 2004 -0.092 0.091

2005 -0.058 0.214 HGV 2005 -0.056 0.225 2005 0.113 0.023 Aggriculture 2006 -0.020 0.429 2006 0.066 0.115 loweducated 2006 -0.086 0.110 2007 0.060 0.122 2007 0.034 0.244 2007 0.028 0.276 2008 -0.012 0.481 2008 0.000 0.440 2008 0.000 0.442 2009 -0.036 0.334 2009 -0.022 0.418 2009 0.000 0.444 2000 -0.013 0.473 2000 -0.130 0.026 2000 -0.024 0.404 2001 0.024 0.296 2001 -0.077 0.138 2001 0.022 0.302 2002 -0.068 0.173 2002 -0.047 0.274 2002 0.080 0.065 2003 -0.078 0.134 2003 -0.040 0.306 2003 0.008 0.387 2004 -0.092 0.091 2004 -0.153 0.011 2004 -0.053 0.240

2005 0.113 0.023 FVU 2005 -0.044 0.287 2005 0.002 0.428 Construction

2006 -0.086 0.110 2006 -0.071 0.159 higheducated 2006 0.006 0.404 2007 0.028 0.276 2007 0.079 0.069 2007 -0.112 0.049 2008 0.000 0.442 2008 0.009 0.386 2008 0.004 0.416 2009 0.000 0.444 2009 0.046 0.185 2009 0.110 0.028 2000 -0.024 0.404 2000 -0.012 0.480 2000 -0.012 0.480 2001 0.022 0.302 2001 0.054 0.155 2001 0.054 0.155 2002 0.080 0.065 2002 -0.106 0.048 2002 -0.106 0.048 2003 0.008 0.387 2003 -0.004 0.468 2003 -0.004 0.468 2004 -0.053 0.240 2004 -0.036 0.334 2004 -0.036 0.334

2005 0.002 0.428 OEPR 2005 0.032 0.255 2005 0.032 0.255

Production 2006 0.006 0.404 2006 -0.087 0.104 2006 -0.087 0.104 parttim worker 2007 -0.112 0.049 2007 0.008 0.392 2007 0.008 0.392 2008 0.004 0.416 2008 -0.007 0.485 2008 -0.007 0.485 2009 0.110 0.028 2009 -0.063 0.196 2009 -0.063 0.196 * p-values below 0.05 or below 0.1 are highlighted

27 Appendix F Model 6.1a estimated with random effects

Variable Coeff. t-Value AGR_west 0.01219 4.93 *** AGR_west_vp 0.00854 4.01 *** Agriculture AGR_east 0.00956 3.56 *** AGR_east_vp -0.00309 -1.22 CONST_west 0.02464 3.22 *** CONST_west_vp 0.03792 6.11 *** Construction CONST_east 0.06799 11.17 *** CONST_east_vp 0.06069 6.84 *** PROD_west 0.01458 3.33 *** PROD_west_vp 0.01103 2.00 ** Production PROD_east 0.02011 2.54 ** PROD_east_vp 0.02138 2.91 *** THT_west 0.08656 7.57 *** Trade, THE_west_vp 0.01221 1.05 Hospitality, THT_east 0.00151 0.10 Transportation THT_east_vp -0.05348 -2.51 ** FIS_west 0.01064 1.06 Finance, FIS_west_vp 0.00664 0.95 Insurance, FIS_east 0.01815 0.98 Services FIS_east_vp 0.00046 0.04 PPS_west 0.04087 1.89 * Public and PPS_west_vp -0.01407 -1.11 private services PPS_east 0.14898 5.49 *** PPS_east_vp 0.03067 1.12 crisis west 0.48614 4.08 *** Crisis crisis east 0.22990 1.07 gwage_ -0.06596 -2.15 ** Factor Prices interst rate 0.67711 8.36 *** _cons 0.10275 1.70 * Observations = 1090 R² = 0.5823 F(28,108) = 155.85 corr(u_i, Xb) = 0.1235

***, **, *: Significant at 1, 5, 10%-level

28 Appendix G Model 6.1b estimated with fixed effects

Random effects Trade, Hospitality, Finance, Insurance, Public and private Agriculture Construction Production Transportation Services services

Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value AGR_west 0.07712 7.68 *** AGR_west_vp 0.04474 5.02 *** AGR_east 0.07416 6.3 *** AGR_east_vp 0.01798 1.51 CONST_west 0.22121 8.37 *** CONST_west_vp 0.08609 60.5 *** CONST_east 0.41292 13.19 *** CONST_east_vp 0.27109 6.44 *** PROD_west 0.08121 4.68 PROD_west_vp 0.06348 4.5 PROD_east 0.07641 3.78 PROD_east_vp 0.10295 6.14 THT_west 0.28009 1.85 *** THT_west_vp 0.11805 6.58 *** THT_east 0.14762 6.46 *** THT_east_vp 0.04710 2.36 ** FIS_west 0.01057 0.4 FIS_west_vp 0.14768 4.54 *** FIS_east -0.00487 -0.16 FIS_east_vp 0.07817 2.4 ** PPS_west 0.00371 -1.05 PPS_west_vp 0.22745 2.38 *** PPS_east -0.27044 -4.1 *** PPS_east_vp 0.41327 12.93 *** crisiswest 2.49942 6.23 *** 1.08570 3.70 *** 1.84060 8.96 *** 1.05663 12.22 *** 1.23400 4.81 *** 0.21690 2.25 crisisost 1.88852 3.33 *** 0.59309 0.11 2.01868 5.26 *** 0.81970 5.22 *** 1.98415 4.88 *** -0.14845 1.28 gwage_sectors -0.05400 -0.62 -0.03524 -0.44 -0.02004 -0.24 -0.44307 -9.44 *** 1.35856 14.10 *** 0.12024 0.73 interest rate 0.34190 0.81 1.42165 7.52 *** 1.73127 14.21 *** 0.87240 8.52 *** -0.20244 -1.14 0.27682 -1.26 *** _cons -1.36131 -9.47 *** -1.83707 -14.79 *** -1.06576 -5.81 *** 0.14539 2.67 *** 0.13176 0.59 0.33423 2.30 ** Observations 1090 1090 1090 1090 1090 1090 R² 0.1377 0.5293 0.4187 0.4449 0.1392 0.1811 Wald chi2 (8) 287.37 156.34 92.72 678.22 278.31 237.8

29 Appendix H Model 6.2 estimated with fixed effects

foreign workers low educated workers high educated worker part-time workes Variable (fixed effects) (fixed effects) (random effects) (fixed effects) Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value AGR_west 0.02352 2.06 ** 0.00293 2.29 ** -0.01401 -1.87 * -0.03880 -4.40 *** AGR_west_vp 0.01715 1.74 * 0.00145 1.29 0.01209 1.36 -0.04954 -6.30 *** Agriculture AGR_east 0.03198 1.05 0.00757 7.22 *** -0.01363 -2.33 ** -0.02863 -2.04 ** AGR_east_vp 0.03232 0.59 0.00241 2.08 ** 0.00139 0.22 -0.03876 -2.88 *** CONST_west 0.11808 3.51 *** 0.00071 0.25 0.07185 4.52 *** 0.06846 2.93 *** CONST_west_vp 0.04761 2.00 ** 0.00478 1.89 * 0.02671 1.54 -0.06048 -3.30 *** Construction CONST_east 0.50530 4.35 *** 0.02846 6.90 *** 0.11301 5.32 *** 0.15697 4.43 *** CONST_east_vp -0.36897 -1.90 * 0.01217 3.25 *** 0.07116 2.46 ** 0.10864 2.39 ** PROD_west 0.02844 1.34 -0.00301 -1.74 * -0.00554 -0.46 0.04314 2.55 ** PROD_west_vp 0.03912 1.89 * 0.00394 2.45 ** 0.02377 2.12 ** 0.04894 3.63 *** Production PROD_east -0.02067 -0.28 -0.00430 -1.67 * -0.00612 -0.35 0.06623 2.02 ** PROD_east_vp 0.17097 1.98 ** -0.00218 -0.97 -0.00444 -0.28 0.00160 0.06 THT_west 0.09718 2.35 ** 0.00033 0.07 0.03769 1.65 -0.13054 -4.33 *** Trade, THT_west_vp 0.04899 1.31 0.03087 6.03 *** 0.04960 1.52 -0.12131 -3.00 *** Hospitality, THT_east -0.03435 -0.11 -0.03178 -3.56 *** -0.05523 -1.76 * -0.00502 -0.06 Transportation THT_east_vp 0.33812 1.70 * -0.00387 -0.30 -0.06206 -1.14 -0.03160 -0.48 FIS_west 0.03445 0.81 0.00848 1.70 * 0.01057 0.43 0.07831 2.64 *** Finance, FIS_west_vp -0.09669 -2.77 *** 0.00044 0.09 -0.01519 -0.69 0.04661 1.51 Insurance, FIS_east 0.16423 0.82 -0.00335 -0.51 -0.00131 -0.04 0.15419 2.40 ** Services FIS_east_vp 0.30898 1.45 0.00106 0.19 -0.05286 -1.46 0.05409 0.74 PPS_west -0.02498 -0.42 0.01123 1.56 0.02685 0.68 0.12164 1.92 * Public and PPS_west_vp -0.10971 -2.26 ** 0.01370 2.19 ** 0.06627 1.75 * -0.19111 -3.24 *** private PPS_east 0.13574 0.33 0.00397 0.30 -0.22999 -1.70 * -0.01182 -0.10 services PPS_east_vp -0.40522 -1.99 ** 0.01346 1.15 -0.04732 -0.79 0.02235 0.17 crisiswest 4.23801 8.00 *** 0.92571 14.51 *** 1.55424 5.39 *** 3.69227 12.85 *** Crisis crisisost 7.86963 5.97 *** 1.22982 8.82 *** 1.79487 3.04 *** 5.89865 5.92 *** gwage_ 0.05893 0.31 0.10298 6.68 *** -0.04763 -0.66 -0.67153 -8.00 *** Factor Prices interest rate 3.07160 5.48 *** 0.13164 3.82 *** 1.15951 6.07 *** 1.29221 11.08 *** _cons -0.94667 -2.29 ** 13.23269 54.30 *** 2.59405 15.68 *** 4.47987 16.15 *** Observation 1090 1090 1090 1090 R² 0.3095 0.6576 0.5021 0.3462 F(28.108) 32.16 1275.55 21.51 853.81 ***, **, *: Significant at 1, 5, 10%-level

30 Appendix I

Variance inflation factor (VIF), estimation part-time employment.

Variable VIF 1/VIF

crisisost 2.68 0.373243 ostbau_p 2.35 0.425565 g_lt_ir_ 2.17 0.460175 crisiswest 2.10 0.475850 ostoepr 2.01 0.498326 ostoepr_p 1.98 0.503951 ostbau 1.81 0.551282 westprod 1.67 0.598600 ostfvu_p 1.63 0.611820 ostprod 1.58 0.631966 westland 1.58 0.631986 ostfvu 1.57 0.637992 westhgv 1.47 0.682090 osthgv 1.42 0.705349 ostland_p 1.40 0.713660 osthgv_p 1.40 0.714436 ostland 1.36 0.733668 westoepr 1.34 0.748465 westland_p 1.33 0.753086 westbau_p 1.33 0.753242 westbau 1.33 0.754156 westhgv_p 1.23 0.813958 ostprod_p 1.22 0.817713 gwage_ 1.22 0.821177 westfvu_p 1.20 0.833632 westoepr_p 1.19 0.840311 westfvu 1.17 0.856592 westprod_p 1.12 0.893694

Mean VIF 1.57

31 Appendix J Labor Market District Deggendorf Regen Fulda Hersfeld-Rotenburg Aachen Aachen Deggendorf Straubing Fulda Wartburgkreis Aachen Aachen Deggendorf Straubing-Bogen Garmisch- Garmisch-Partenkirchen Aachen Düren Dessau Anhalt-Bitterfeld Partenkirchen Aachen Heinsberg Dessau Dessau-Roßlau Gera Gera Ahrweiler Ahrweiler Dessau Wittenberg Gera Greiz Altenburger Land Altenburger Land Dithmarschen Dithmarschen Gera Jena Altötting Altötting Dresden Bautzen Gera Saale-Holzland-Kreis Altötting Mühldorf a. Inn Dresden Dresden Gera Saale-Orla-Kreis Altötting Rottal-Inn Dresden Görlitz Gera Saalfeld-Rudolstadt Amberg Amberg Dresden Meißen Gießen Gießen Amberg Amberg-Sulzbach Sächsische Schweiz- Gießen Lahn-Dill-Kreis Dresden Ansbach Ansbach Osterzgebirge Gießen Marburg-Biedenkopf Ansbach Ansbach Düsseldorf Duisburg Goslar Goslar Weißenburg- Düsseldorf Düsseldorf Göttingen Eichsfeld Ansbach Gunzenhausen Düsseldorf Kleve Göttingen Göttingen Aschaffenburg Aschaffenburg Düsseldorf Krefeld Göttingen Nordhausen Aschaffenburg Aschaffenburg Düsseldorf Mettmann Göttingen Northeim Aschaffenburg Miltenberg Düsseldorf Mönchengladbach Göttingen Osterode am Harz Augsburg Aichach-Friedberg Düsseldorf Rhein-Kreis Neuss Greifswald Greifswald Augsburg Augsburg Düsseldorf Solingen Greifswald Ostvorpommern Augsburg Augsburg Düsseldorf Viersen Halle Burgenlandkreis Augsburg Dillingen a.d. Donau Düsseldorf Wesel Halle Halle (Saale) Augsburg Donau-Ries Düsseldorf Wuppertal Halle Mansfeld-Südharz Bad Kreuznach Bad Kreuznach Emden Aurich Halle Saalekreis Bamberg Bamberg Emden Emden Hamburg Hamburg Bamberg Bamberg Emden Leer Hamburg Harburg Bayreuth Bayreuth Emden Wittmund Hamburg Herzogtum Lauenburg Bayreuth Bayreuth Emsland Emsland Hamburg Lüneburg Bayreuth Kulmbach Emsland Grafschaft Bentheim Hamburg Pinneberg Berchtesgadener Land Berchtesgadener Land Erfurt Erfurt Hamburg Segeberg Berchtesgadener Land Traunstein Erfurt Gotha Hamburg Stade Berlin Barnim Erfurt Ilm-Kreis Hamburg Stormarn Berlin Berlin Erfurt Kyffhäuserkreis Hamm Hamm Brandenburg an der Erfurt Sömmerda Hannover Celle Berlin Havel Erfurt Unstrut-Hainich-Kreis Hannover Hameln-Pyrmont Berlin Dahme-Spreewald Essen Bochum Hannover Hildesheim Berlin Frankfurt (Oder) Essen Bottrop Hannover Region Hannover Berlin Havelland Essen Dortmund Hannover Schaumburg Berlin Märkisch-Oderland Essen Ennepe-Ruhr-Kreis Heilbronn Heilbronn Berlin Oberhavel Essen Essen Heilbronn Heilbronn Berlin Oder-Spree Essen Gelsenkirchen Heilbronn Hohenlohekreis Berlin Potsdam Essen Hagen Heilbronn Neckar-Odenwald-Kreis Berlin Potsdam-Mittelmark Essen Herne Hof Hof Berlin Teltow-Fläming Essen Recklinghausen Hof Hof Birkenfeld Birkenfeld Essen Unna Holzminden Holzminden Bremen Bremen Flensburg Flensburg Holzminden Höxter Bremen Delmenhorst Flensburg Nordfriesland Ingolstadt Eichstätt Bremen Diepholz Flensburg Schleswig-Flensburg Ingolstadt Ingolstadt Bremen Osterholz Frankfurt am Main Darmstadt Neuburg- Ingolstadt Bremen Rotenburg (Wümme) Frankfurt am Main Darmstadt-Dieburg Schrobenhausen Bremen Verden Frankfurt am Main Frankfurt am Main Ingolstadt Pfaffenhofen a.d. Ilm Bremerhaven Bremerhaven Frankfurt am Main Groß-Gerau Kaiserslautern Kaiserslautern Bremerhaven Cuxhaven Frankfurt am Main Hochtaunuskreis Kaiserslautern Kaiserslautern Calw Calw Frankfurt am Main Main-Kinzig-Kreis Kaiserslautern Kusel Chemnitz Chemnitz Frankfurt am Main Main-Taunus-Kreis Karlsruhe Baden-Baden Chemnitz Erzgebirgskreis Frankfurt am Main Mainz Karlsruhe Enzkreis Chemnitz Mittelsachsen Frankfurt am Main Mainz-Bingen Karlsruhe Karlsruhe Chemnitz Vogtlandkreis Frankfurt am Main Offenbach Karlsruhe Karlsruhe Chemnitz Zwickau Frankfurt am Main Offenbach am Main Karlsruhe Pforzheim Coburg Coburg Frankfurt am Main Rheingau-Taunus-Kreis Karlsruhe Rastatt Coburg Coburg Frankfurt am Main Wetteraukreis Kassel Kassel Coburg Kronach Frankfurt am Main Wiesbaden Kassel Kassel Coburg Lichtenfels Breisgau- Kassel Schwalm-Eder-Kreis Freiburg im Breisgau Coburg Sonneberg Hochschwarzwald Kassel Waldeck-Frankenberg Cottbus Cottbus Freiburg im Breisgau Emmendingen Kassel Werra-Meißner-Kreis Cottbus Elbe-Elster Freiburg im Breisgau Freiburg im Breisgau Kaufbeuren Kaufbeuren Cottbus Oberspreewald-Lausitz Freiburg im Breisgau Ortenaukreis Kaufbeuren Ostallgäu Cottbus Spree-Neiße Freudenstadt Freudenstadt Kempten Kempten (Allgäu) Deggendorf Deggendorf Fulda Eisenach Kempten Oberallgäu Deggendorf Dingolfing-Landau Fulda Fulda Kiel Kiel

32 Kiel Neumünster Nienburg (Weser) Nienburg (Weser) Schwerin Ludwigslust Kiel Plön Nürnberg Erlangen Schwerin Nordwestmecklenburg Kiel Rendsburg-Eckernförde Nürnberg Erlangen-Höchstadt Schwerin Parchim Koblenz Cochem-Zell Nürnberg Forchheim Schwerin Schwerin Koblenz Koblenz Nürnberg Fürth Schwerin Wismar Koblenz Mayen-Koblenz Nürnberg Fürth Altenkirchen Siegen Koblenz Neuwied Nürnberg Neumarkt i.d. OPf. (Westerwald) Koblenz Rhein-Hunsrück-Kreis Neustadt a.d. Aisch-Bad Siegen Olpe Nürnberg Koblenz Rhein-Lahn-Kreis Windsheim Siegen Siegen-Wittgenstein Koblenz Westerwaldkreis Nürnberg Nürnberg Sigmaringen Sigmaringen Köln Bonn Nürnberg Nürnberger Land Sigmaringen Zollernalbkreis Köln Euskirchen Nürnberg Roth Soest Hochsauerlandkreis Köln Köln Nürnberg Schwabach Soest Märkischer Kreis Köln Leverkusen Oberhausen Mülheim an der Ruhr Soest Soest Köln Oberbergischer Kreis Oberhausen Oberhausen Soltau-Fallingbostel Soltau-Fallingbostel Köln Rhein-Erft-Kreis Odenwaldkreis Odenwaldkreis Steinburg Steinburg Rheinisch-Bergischer Oldenburg Ammerland Stendal Altmarkkreis Salzwedel Köln Kreis Oldenburg Cloppenburg Stendal Lüchow-Dannenberg Köln Rhein-Sieg-Kreis Oldenburg Oldenburg Stendal Stendal Landau in der Pfalz Germersheim Oldenburg Oldenburg (Oldenburg) Stendal Uelzen Landau in der Pfalz Landau in der Pfalz Oldenburg Vechta Stralsund Nordvorpommern Landau in der Pfalz Südliche Weinstraße Oldenburg Wesermarsch Stralsund Stralsund Landsberg am Lech Landsberg am Lech Ostablkreis Heidenheim Stuttgart Böblingen Landshut Landshut Ostablkreis Ostalbkreis Stuttgart Esslingen Landshut Landshut Ostablkreis Schwäbisch Hall Stuttgart Göppingen Leipzig Leipzig Paderborn Bielefeld Stuttgart Ludwigsburg Leipzig Leipzig Paderborn Gütersloh Stuttgart Rems-Murr-Kreis Leipzig Nordsachsen Paderborn Herford Stuttgart Reutlingen Limburg-Weilburg Limburg-Weilburg Paderborn Lippe Stuttgart Stuttgart Lörrach Lörrach Paderborn Minden-Lübbecke Stuttgart Tübingen Lörrach Waldshut Paderborn Paderborn Suhl Hildburghausen Lübeck Lübeck Passau Freyung-Grafenau Schmalkalden- Suhl Lübeck Ostholstein Passau Passau Meiningen Magdeburg Börde Passau Passau Suhl Suhl Magdeburg Harz Pirmasens Pirmasens Trier Bernkastel-Wittlich Magdeburg Jerichower Land Pirmasens Südwestpfalz Trier Eifelkreis-Bitburg-Prüm Magdeburg Magdeburg Pirmasens Zweibrücken Trier Trier Magdeburg Salzlandkreis Prignitz Ostprignitz-Ruppin Trier Trier-Saarburg Mannheim Bad Dürkheim Prignitz Prignitz Trier Vulkaneifel Mannheim Bergstraße Ravensburg Bodenseekreis Uckermark Uckermark Mannheim Frankenthal (Pfalz) Ravensburg Konstanz Uckermark Uecker-Randow Mannheim Heidelberg Ravensburg Lindau (Bodensee) Ulm Alb-Donau-Kreis Mannheim Ludwigshafen am Rhein Ravensburg Ravensburg Ulm Biberach Mannheim Mannheim Regensburg Cham Ulm Günzburg Mannheim Rhein-Neckar-Kreis Regensburg Kelheim Ulm Neu-Ulm Mannheim Rhein-Pfalz-Kreis Regensburg Regensburg Ulm Ulm Mannheim Speyer Regensburg Regensburg Vogelsbergkreis Vogelsbergkreis Memmingen Memmingen Regensburg Schwandorf Weiden i.d. Opf. Neustadt a.d. Waldnaab Memmingen Unterallgäu Remscheid Remscheid Weiden i.d. Opf. Tirschenreuth Bad Tölz- Rosenheim Rosenheim Weiden i.d. Opf. Weiden i.d. OPf. München Wolfratshausen Rosenheim Rosenheim Wunsiedel i. Weiden i.d. Opf. München Dachau Rostock Bad Doberan Fichtelgebirge München Ebersberg Rostock Güstrow Weilheim-Schongau Weilheim-Schongau München Erding Rostock Rostock Weimar Weimar München Freising Rügen Rügen Weimar Weimarer Land München Fürstenfeldbruck Saarbrücken Merzig-Wadern Wilhelmshaven Friesland München Miesbach Saarbrücken Neunkirchen Wilhelmshaven Wilhelmshaven München München Saarbrücken Saarlouis Wolfsburg Braunschweig München München Saarbrücken Saarpfalz-Kreis Wolfsburg Gifhorn München Starnberg Saarbrücken St. Wendel Wolfsburg Helmstedt Münster Borken Stadtverband Wolfsburg Peine Saarbrücken Münster Coesfeld Saarbrücken Wolfsburg Salzgitter Münster Münster Schwarzwald-Baar- Wolfsburg Wolfenbüttel Rottweil Münster Osnabrück Kreis Wolfsburg Wolfsburg Münster Osnabrück Schwarzwald-Baar- Worms Alzey-Worms Schwarzwald-Baar-Kreis Münster Steinfurt Kreis Worms Donnersbergkreis Schwarzwald-Baar- Worms Worms Münster Warendorf Tuttlingen Neubrandenburg Demmin Kreis Würzburg Kitzingen Neubrandenburg Mecklenburg-Strelitz Schweinfurt Bad Kissingen Würzburg Main-Spessart Neubrandenburg Müritz Schweinfurt Haßberge Würzburg Main-Tauber-Kreis Neubrandenburg Neubrandenburg Schweinfurt Rhön-Grabfeld Würzburg Würzburg Neustadt an der Neustadt an der Schweinfurt Schweinfurt Würzburg Würzburg Weinstraße Weinstraße Schweinfurt Schweinfurt

33